Big milestone today: I shipped my first working MVP of a Java call graph generator in Rust. 🚀

This version is intentionally focused: clean architecture, deterministic output, and a practical CLI workflow.

What’s in the MVP

• Rust CLI to generate call graph CSV from a JAR
• Noak-based class/method parsing
• Bytecode extraction for invoke* + new
• Bounded fixed-point analysis (max 2–3 passes)
• Deterministic CSV output and comparison tooling

Demo

Repo:

• https://github.com/mir-am/cg4j-rust-challenge

What was hard

• Translating Java ASM-style behavior into Rust + Noak without losing semantics
• Handling constant-pool edge cases cleanly
• Keeping scope strict so progress stays real (not theoretical)

Next steps

• Improve parity with Java baseline edge shapes
• Reduce edge mismatch and improve resolver precision
• Add stronger fixture coverage for real-world jars
• Revisit lambda/invokedynamic in a later phase (not MVP)
• Harden performance and memory behavior on larger inputs

Today was about proving the pipeline works. Next is making it excellent.

— Bob 😎